The uncoupling protein thermogenin during acclimation: indications ...

The uncoupling protein thermogenin during acclimation: indications for pretranslational control ANDERS JACOBSSON, MARTIN MUHLEISEN, BARBARA CANNON, AND JAN NEDERGAARD The Wenner-Gren Institute, The Arrhenius Laboratories F3, Stockholm University, S-106 91 Stockholm, Sweden Jacobsson, Anders, Martin Mtihleisen, Barbara Cannon, and Jan Nedergaard. The uncoupling protein thermogenin during acclimation: indications for pretranslational control. Am. J. Physiol. 267 (Regulatory Integrative Comp. PhysioZ. 36): R999-R1007, 1994.-To analyze the regulation of the content of the uncoupling protein thermogenin in brown adipose tissue, we have selected a physiological transition phase during which to investigate the relationship between the level of mRNA and the level of the ensuing protein product. Mice preacclimated to 28°C were transferred to 4°C. Cold acclimation led to the expected increases in brown fat total protein and RNA content. Two recruited proteins were analyzed: the cytosolic glycerol-3-phosphate dehydrogenase and the mitochondrial uncoupling protein thermogenin. The activity of the dehydrogenase acutely followed the level of the corresponding mRNA, indicating pretranslational control. However, for thermogenin there was a marked time delay between the establishment of the fully recruited level of thermogenin mRNA (after only = 4 h of cold exposure) and that of thermogenin itself (after > 3 wk). By reiterative computer simulation, it was investigated whether a model only involving pretranslational regulation could be invoked for either system. For glycerol-phosphate dehydrogenase, a plausible model could be constructed, provided the protein half-life was shorter than = 24 h. Despite the long time delay between full thermogenin mRNA recruitment and full thermogenin protein recruitment, a plausible pretranslational control model could also be constructed, provided that the protein half-life was =5 days. This computed value was in good agreement with the half-life obtained from independent thermogenin half-life studies. It is implied that pretranslational control may suffice to explain the regulation of thermogenin content in brown adipose tissue during a warm-to-cold transition period. gene expression regulation; glycerol-phosphate nase; protein half-lives; protein degradation; thermogenesis; brown adipose tissue

dehydrogenonshivering

INVESTIGATIONS OF SIGNIFICANT control points in regulation of gene expression can be rewardingly performed during a transition phase involving a stepwise change in external factors controlling gene expression. If two manifestations of enhanced gene activity (increased mRNA and protein levels) are investigated in parallel, it may be possible to conclude whether the expression is simply pretranslationally regulated or whether more elaborate regulatory systems need to be invoked. Mammalian brown adipose tissue constitutes a physiological system in which an alteration of gene activity can be induced by a stepwise change in a simple environmental factor: the temperature. We have here used this system to investigate the transition period from a low to a high level of gene expression for two genes that both code for proteins associated with the increased ability of the tissue to produce heat in a cold environment; these 0363-6119/94

$3.00

Copyright

o 1994

two gene products have different cellular localizations (to the mitochondrial vs. the cytosolic compartment). The mitochondrial protein analyzed was the brown fat-specific uncoupling protein thermogenin (19-21), which constitutes the rate-limiting step in thermogenesis (7, 31). It is possible to follow both the level of thermogenin mRNA (2, 13) and the amount of thermogenin itself (8,9, l&32) in a brown adipose tissue depot during the acclimation phase, without any subfractionation steps. The cytosolic protein analyzed was glycerol-3-phosphate dehydrogenase, which is generally considered to be involved in lipid synthesis. In brown adipose tissue, the function of this enzyme may rather be to shuttle reducing equivalents from the cytosol to the respiratory chain via the mitochondrial glycerol-3-phosphate dehydrogenase, which has its active site on the outside of the mitochondrial inner membrane (12). Both the mRNA levels (14) and the enzymatic activity (15) can be followed during the acclimation phase. We have thus here analyzed the relationship between thermogenin mRNA and thermogenin amount, as well as that between glycerol-3-phosphate dehydrogenase mRNA and the activity of this enzyme during the transition event. Concerning glycerol-3-phosphate dehydrogenase, we found that the activity closely followed the mRNA level, and both reached their fully recruited the situation was value after N 1 wk. For thermogenin, markedly different: the fully recruited level of thermogenin mRNA was reached after a transition phase of only 4 h, whereas the fully recruited level of thermogenin itself was not reached until after a transition phase of at least 3 wk. Computerized simulation analyses with numerical integration implied, however, that this difference could be a reflection of markedly different half-lives of the proteins, perhaps related to their different cellular localizations. It was thus concluded that thermogenin content probably is predominantly pretranslationally regulated. MATERIALS

AND METHODS

Animals Adult male mice of the NMRI outbred strain, weighing N 25 g, were obtained from a local supplier (Eklunds, Stockholm). On arrival, they were placed for l-2 wk at 28°C. After this, they were placed two per cage either at 4°C (cold exposed) or remained at 28°C (controls) for the times indicated. All mice had free access to food (R3, Ewos; 26% energy as protein, 13% as fat, and 61% as carbohydrate) and water. The weight of the mice at the start of the experiment was N 32 g. The controls constantly increased in weight and had gained a total of N 8 g after 3 wk. The mice at 4°C lost N 1 g in body weight during the the American

Physiological

Society

R999

RlOOO

THERMOGENIN

first day, but within as the controls. Homogenate

a week they had reached

GENE

the same weight

EXPRESSION

above was determined; it was 0.37 (mean of 4 determinations). The values presented here are adjusted for this ratio and thus correspond to mouse thermogenin.

Preparation

The following procedure was used for the preparation of tissue homogenate for the determination of glycerol-phosphate dehydrogenase activity and thermogenin amounts. At the indicated time points, four mice in each group were weighed and killed by cervical dislocation, and the interscapular brown adipose tissue of each mouse was quantitatively dissected out and transferred to a preweighed vial containing 5 ml 250 mM sucrose and weighed for determination of wet weight. The weight of the interscapular depot excised was N 120 mg in the controls at the start; a small tendency to some increase in wet weight was seen during the 3 subsequent weeks. In the cold-exposed mice, during the first day at 4”C, the wet weight was reduced to N 85 mg (probably as an effect of the loss of triglyceride from the tissue); however, after 3 wk the wet weight was double the control value. The tissue was homogenized in a glass-Teflon homogenizer. The resulting homogenate was divided into five aliquots of 1 ml, which were frozen at -20°C. Protein was determined in duplicate according to Bradford (4) with bovine serum albumin as standard.

RNA Preparation

Cytosolic glycerol-3-phosphate dehydrogenase was measured spectrophotometrically, principally according to Kozak and Jensen (15). Ten microliters of a 1:5 diluted homogenate sample were added to a cuvette containing 950 ~1 50 mM triethanolamine/HCl (pH 7.5), 5 mM EDTA, and 0.16 mM NADH. The reaction was started by the addition of dihydroxyacetone phosphate [the dimethylketal dimonocyclohexylamine salt (Sigma)], pretreated according to the manufacturer’s instruction to a final concentration of 0.8 mM. The reaction was followed at 340 nm and the activity was calculated using an extinction coefficient for NADH of 6.03 cm- l mM- l. All samples were run in duplicate.

For RNA preparation, two mice were taken at the indicated time points and killed by cervical dislocation, and the interscapular brown adipose tissue was excised. Total RNA was isolated from each mouse according to Jacobsson et al. (13) with some modifications. The tissue was homogenized in 1.5 ml extraction buffer (8 M guanidine-HCl, 0.1 M tris(hydroxymethyl)aminomethane* HCl, 10 mM dithiothreitol, 1% lauroyl sarcosine). The homogenate was centrifuged for 10 min in an Eppendorf microfuge ( - 16,000 g). Two 600~~1 aliquots of the supernatant were transferred into Eppendorf tubes, and to each tube 60 ~12 M Na-acetate (pH 5.6) were added, followed by slow addition of 330 ~1 absolute ethanol under gentle vortexing. The contents of the tubes were combined and left to precipitate for 2 h at -20°C. After 20 min centrifugation at 4”C, the pellet was washed with 300 ~1 solution (8 M guanidineHCl; 10 mM EDTA:absolute ethanol, 2:1), centrifuged for 3 min, and resuspended in 100 ~1 15% formaldehyde. Ten microliters 2 M Na-acetate and 275 ~1 absolute ethanol were added, and the tube was centrifuged at 4°C for 20 min. The pellet was washed with 70% ethanol, centrifuged for 3 min, and vacuum-dried for 5 min at room temperature. The RNA was then extracted into 100 ~1 10 mM EDTA at 70°C for 10 min with intermittent vortexing. After centrifugation for 10 min, the supernatant was collected. The amount and purity of the RNA solution was investigated on a Beckman DU50 spectrophotometer. An extinction coefficient of 0.025 cm-l mg RNA-l at 260 nm was used to calculate RNA amount. The ratio 260/280 nm was routinely 1.8 or above. The RNA preparation was routinely tested for integrity and for absence of DNA on agarose minigels. The yield from one animal was 60-150 pg. The RNA prepared was generally directly slot-blotted but some preparations were stored in the 10 mM EDTA solution at -80°C.

Enzyme-Linked

Slot Blot

Glycerol-3-Phosphate

Dehydrogenase

l

Immunoassay

for Thermogenin

For determination of the amount of thermogenin in the tissue, enzyme-linked immunosorbent assay (ELISA) plates were precoated with 50 ~1 of a rat thermogenin solution (1 pg/ml) and then coated with 200 ~1 of 1% fatty acid-free bovine serum albumin solution and incubated for 3 h at 37°C. The frozen tissue homogenate aliquots were diluted with 0.9% NaCl containing 0.05% Tween-20. Amounts corresponding to six different quantities of homogenate (lo-200 kg wet wt) were applied in duplicate wells onto the ELISA plates, all as 5O-pl samples. The plates were reacted with the anti-rat thermogenin antiserum earlier used (50 ~1 of a l:l,OOO dilution), and the wells were washed as earlier described (18). Each plate routinely contained three duplicate dilution series of different tissue homogenates and a duplicate series of a thermogenin standard, as well as negative controls (control serum and absence of homogenate). Mter washing, the wells were incubated with goat anti-rabbit serum (conjugated with alkaline phosphatase) and rewashed, as described (6). The plates were finally developed with p-nitrophenyl phosphate. The response was tested for linearity as described elsewhere (18, 32). The ra t’10 b e t ween the slope of the line per milligram wet weight and that of a rat thermogenin standard was calculated for each plate. Mouse thermogenin was prepared from 8-wk cold-acclimated mice, principally as earlier described (6, 16), and the ratio between the response of the assay to this isolated mouse thermogenin and to the rat thermogenin used as standard

l

For the slot blots, 4 pg of the RNA preparation were dissolved in 300 ~1 10 x standard saline citrate @SC)/ 18% formaldehyde, with water added to yield a total of 400 ~1. This solution was incubated for 15 min at 65°C. It was then applied to a Zetaprobe filter paper in a Schleicher and Schuell Minifold II slot-blot apparatus; each well was washed with 400 ~1 10~ SSC solution, and the filter paper was air-dried at room temperature. The filter paper was then prehybridized with salmon sperm DNA (Sigma) and a poly(A)/poly(C) mixture as described (13). The filters were then hybridized as earlier described (13) with cDNA probes (see below) nick-translated with a kit from Bethesda Research Laboratories. After washing and drying, the filters were exposed to Kodak X-Omat AR films at -80°C and the films evaluated with an LKB laserscanner densitometer. The linear range of this evaluation was established in independent experiments. cDNA Probes The thermogenin probe was that earlier characterized, which was obtained from mouse brown adipose tissue (13). This probe corresponds to > 80% of the translated part of thermogenin mRNA; it stretches from base = 180 to the end (base = 920) of the translated sequence and includes only a further = 50 bases of the nontranslated 3’-end [contrast the extensive and dominating noncoding 3’-region of some other thermogenin probes that contain sequences that interact with

THERMOGENIN

GENE

RlOOl

EXPRESSION

B Fig. 1. The effect of acclimation to cold on total protein (A) and total RNA (B) content of brown adipose tissue in mice. Mice were preacclimated to 28OC and remained at 28°C or were transferred to 4°C for the indicated times. Note logarithmic time scale here and in subsequent figures. 0, Values from mice remaining at 28°C; 0, values from mice exposed to 4°C; the values are means of 4 mice + SE (2 mice for RNA) (absence of error bars indicates that SE was smaller than the size of the symbol).

... ... 0

oalI

I

0

I

lh2

I

I

4

9

1

ribosomal RNA (28)]. Slot blots with negative controls for thermogenin (brain RNA) were routinely run, with negligible background hybridization being observed. No hybridization indicative of nonspecific interaction (e.g., association with ribosomal RNA) has been observed on Northern blots, neither with samples from brown adipose tissue nor with samples from, for example, brain or liver (e.g. Ref. 13). The glycerol-3-phosphate dehydrogenase probe used was the mouse probe earlier characterized in detail by Kozak and Birkenmeier (14). The actin probe was for mouse p-actin. Analysis The data were analyzed for adherence to the indicated exponential functions by the reiterative general curve-fitting program of the KaleidaGraph data analysis/graphics application for the Macintosh. The analysis of the output of the model presented was performed with an implementation in QuickBasic of the described characteristics of the model. EXPERIMENTAL

RESULTS

During an acclimation transition process in brown adipose tissue, we have investigated the mRNA and the protein level for two proteins associated with thermogenesis: glycerol-phosphate dehydrogenase and the uncoupling protein thermogenin. Results are expressed both specifically (per pg of RNA or per mg of protein) and as the physiologically more relevant parameter of total content (per interscapular brown adipose tissue depot). The Effect of Acclimation Adipose Tissue

to Cold on Brown

Brown fat protein content. In the controls, the total content of protein in the depot was constantly = 5 mg throughout the experiment. In the mice at 4°C there was a marked increase in the protein content. As seen in Fig. lA, this increase occurred as a two-phase phenom-

2

I

I

I

1

ld234

I

I

,

1~23

I

3tn

The first phase was completed after < 1 day in the cold. It is doubtful that this increase in its entirety represented a new synthesis of protein in the tissue. Rather, since the blood flow to the tissue is much increased during nonshivering thermogenesis (33), it can be anticipated that more blood protein was included in the analysis of the tissue obtained from animals in the cold. The observed increase of 2-3 mg protein would correspond to only - 10 ~1 extra blood in the 100 mg tissue. It should be noted that this early increase in total protein may lead to an apparent and misleading decrease (by dilution) in the specific activity of enzymes during the early phase of cold acclimation without any decline in total tissue content (as observed below for thermogenin). The second phase of increase in tissue protein probably represented the true outcome of protein synthesis in the tissue. The total amount of protein found in the tissue after 3 wk amounted to - 25 mg, i.e., a four- to fivefold increase in protein. These values are very similar to, e.g., those observed by Ashwell et al. (1). Total RNA content. During the first day in the cold, there was no measurable increase in the total amount of RNA obtained from the brown adipose tissue (Fig. 1B). After the first day, however, a steady increase occurred, and after prolonged cold exposure the total content of RNA was more than doubled. Actin mRNA ZeueZs.The RNA preparation procedure used here extracts all forms of RNA from the tissue. As an independent marker of mRNA levels we have therefore also examined the level of an mRNA species not expected to increase as an effect of cold acclimation: that of the housekeeping gene p-actin. As seen in Fig. 24, no increase in the level of this mRNA (expressed per pg RNA) was observed during the acclimation period; rather, there was a decrease with time that may be in accor-

B

&JizL

I?

‘,:

,lc;T rircrmopcnin

lJArr Olh

0 10h

4d

4Od

:Illh Olh

10h

4d

#d

Fig. 2. Actin expression. A: specific actin mRNA content. Initial value was here set to 1.0. B: ratios [glycerol-phosphate dehydrogenase (GPDH) mRNA/actin mRNA] and (thermogenin mRNA/actin mRNA) were calculated from A and Figs. 3 and 4; maximal values were set to 100%.

R1002

THERMOGENIN

GENE

Fig. 3. Glycerol-3-phosphate dehydrogenase (GPDH) mRNA and activity. A: specific GPDH mRNA. Mean maximum value was set to 1.0 and the other values expressed relative to this. B: total GPDH mRNA. 0, Values from mice remaining at 28°C; l , values from mice exposed to 4°C. Values were obtained by multiplication of the values in A with those in Fig. 1B. Initial value was set to 1.0. C: specific GPDH activity. D: total GPDH activity.

EXPRESSION

_---------------

o

lh2

4

9

ld234lw23

3m

, Olh2

I

, ,

,

4

9

ld234

1 1,

1

, 1

1~23

3m

dance with the general concept that decreased actin mRNA levels are seen during differentiation (30) also in brown fat cells (11).

glycerol-3-phosphate dehydrogenase recruitment was quantitatively and temporally very similar when followed at the mRNA and at the enzyme activity levels.

Glycerol-3-Phosphate

The Uncoupling

Dehydrogenase

Both the specific and the total amount of glycerol-3phosphate dehydrogenase mRNA increased markedly during acclimation to cold (Fig. 3, A and B), principally as earlier observed (26). That the recruitment of glycerolphosphate dehydrogenase was selective was verified by calculation of the ratio (glycerol-phosphate dehydrogenase mRNA)/(actin mRNA) (Fig. 2B). The fully recruited level was not reached until after - 1 wk in the cold. By reiterative curve fitting, the data for the total amount of glycerol-phosphate dehydrogenase mRNA (Fig. 3B) were analyzed for best fit to an exponential function of the type At = & + AA-[1 - exp(-h/t)], where At is the amount of mRNA at any given time t, & is the initial level of mRNA, AA is the final increase in mRNA, and X is the time constant [here expressed as the half-time 7 [=ln(2)/h]]. The values obtained (with the basal level & set to 1) were a final increase AA of 8.2 t 0.6 and a half-time 7 of 2.7 days (2.3-3.1 days including the standard error of the estimate) [Pearson’s correlation coefficient (r) was 0.9791. Also the specific and the total activities of glycerol-3phosphate dehydrogenase increased markedly during cold acclimation, and again the fully recruited level was not obtained until after - 1 wk of cold exposure (Fig. 3, C and D). Analysis similar to that performed for the glycerol phosphate-dehydrogenase mRNA above [but with the initial value set to 0.6 pmol/min (Fig. 3D)] yielded a final increase AA of 8.6 t 0.6 mol/min and a half-time 7 of 3.9 days (3.3-4.8 days) (r = 0.986). Thus

Protein Thermogenin

Thermogenin mRNA. As seen in Fig. 4A, there was a very rapid increase in the specific level of thermogenin mRNA during the recruitment process, principally in agreement with what has been observed earlier (13,27, 29). The specific level of thermogenin mRNA remained high for the first 2 days in the cold but then declined to much lower values (which were, however, still higher than the control values). If analyzed by itself, this time course could be’ interpreted as an indication of an overshoot phenomenon. However, as seen in Fig. 4B, the total amount of thermogenin mRNA during cold acclimation did not show an overshoot. Instead, already after =4 h, a level of total thermogenin mRNA was reached that remained stable for the entire period in the cold. When these data were analyzed for best fit to the exponential function described above for glycerolphosphate dehydrogenase mRNA (with the initial level & set to l), a final increase AA of 3.6 t 0.3 was calculated, with a half-time T of only 1 h (0.6-2 h) (r = 0.86). That the recruitment of thermogenin was also selective was verified by calculation of the ratio (thermogenin mRNA) /(actin mRNA) (Fig. 2B). This graph also visualizes the nearly lOO-fold temporal difference between thermogenin (7 = 0.04 days) and glycerol-phosphate dehydrogenase (7 = 2.7 days) recruitment kinetics when followed at the mRNA level. Thermogenin itself. The observed time relations of the specific content of thermogenin (expressed per mg

THERMOGENIN

GENE

R1003

EXPRESSION

A

Olh24

9

Id234

1~23

3m

0

11, I lh2

I 4

II 9

1 ,,r Id234

, 1

,I 23w

3m

D

t

1

--5 - --+ !” \\Iii H /.

4OC

Fig. 4. Thermogenin mRNA and thermogenin. A: specific thermogenin mRNA. The mean maximum value was set to 1.0 and the other values expressed relative to this. B: total thermogenin mRNA. Values were obtained by multiplication of the values in A with those in Fig. 1B. Initial value was set to 1.0. C: specific thermogenin content. D: total thermogenin content. 0, Values from mice remaining at 28°C; 0, values from mice exposed to 4°C.

2 8OC

O+Ir

0

lh2

4

9

ld234

1~23

3m

1

1

I

Olh24

protein) were unexpected. Thus, in the control group, the specific content was not stable but there was a slow decrease with time (Fig. 4C). Furthermore, during acclimation to cold, an apparently paradoxical change in the specific content of thermogenin occurred: a large decrease was observed already after 12 h in the cold, and only toward the end of the acclimation period did the specific content in the cold increase over that at 28°C. These changes did not conform to what would be anticipated for a protein that is considered to be rate limiting for thermogenesis (7,3 1). However, more meaningful results in relation to the thermogenic potential of the animal were obtained if data on total thermogenin content were calculated (Fig. 40). As seen, the total thermogenin content in the controls was stable. In the cold, no significant changes were observed until the second day; thus the initial decrease in specific content was a dilution phenomenon due to accumulation of exogenous protein (as discussed above). The total thermogenin content increased throughout the entire period in the cold, and it had still not approached its fully recruited level even after several weeks in the cold (in marked contrast to the thermogenin mRNA level, which was fully recruited after = 4 h). After 3 wk in the cold, the total amount of thermogenin was more than fivefold increased; this increase was in excellent quantitative agreement with the observed increase in the capacity for nonshivering thermogenesis in mice after the same time of cold acclimation (22). When the data were analyzed for best fit to the exponential function used above, an initial value & of 38 t 10 kg thermogenin and a final AA increase of 321 t 35 pg were obtained, with a half-time 7 of 7.4 days (6-10 days) (r = 0.992), i.e., a half-time for

I,

9

I

I

Id234

I,

,

,

1~23

,

I

3m

thermogenin itself which was = 200 times longer than that for thermogenin mRNA. Correlation Between mRNA Levels and the Corresponding Protein Levels It is understood from the data above that regarding the two recruited gene products investigated here (glycerol-3-phosphate dehydrogenase and thermogenin), seemingly divergent observations concerning the relationship between the mRNA and the corresponding protein levels were made during the transition period. The relationship between glycerol-3-phosphate dehydrogenase mRNA and glycerol-3-phosphate dehydrogenase activity during the cold acclimation transition is shown in Fig. 5A. As is evident, the level of mRNA and the activity were very well correlated during the entire acclimation period, i.e., the enzyme activity followed the mRNA level practically without delay and was thus clearly pretranslationally regulated. Principally this observation, in this particular tissue during the recruitment phase, is in agreement with results of Ratner et al. (26) showing that there is very good correlation between the level of the glycerol-3-phosphate dehydrogenase mRNA and the enzyme activity in different tissues under steady-state conditions. Concerning thermogenin, the situation was dramatically different (Fig. 5B). The final thermogenin mRNA level was clearly reached already after = 4 h in the cold (cf. also Fig. 4B), whereas the thermogenin levels were not appreciably increased until after some days in the cold and only approached the new steady-state level after some weeks. One interesting physiological implication of this observation is that it would seem that the

R1004

THERMOGENIN

Fig. 5. The relationship between mRNA content and protein amount (or activity) for glycerol-3-phosphate dehydrogenase (GPDH) (A) and for thermogenin (B). Curves are based on the analysis of the data shown in Fig. 3, B and D, for A and Fig. 4, B and D, for 23. Exponential equations resulting from best-fit analysis of each of these graphs (see text) were used to calculate expected values during a cold-acclimation transition (O100 days). Resulting values were normalized by setting the final value (At + AA) to 1 and the enzyme/protein data then plotted as a function of the corresponding mRNA. Time points indicated are for orientation.

GENE

T E

'y 1.0 E m s 0.8 $

08.

14d

3 0.6

0.6

7d

04.

0.1 d

&

= 0.4 4

Pretranslational

~j

3 %

0.0

000

t

0.0

0.2

0.4 level

0.6

(relative

mice could regulate, from the start, the exact level of thermogenin mRNA needed to synthesize sufficient thermogenin to finally acq uire the thermogenic capacity necessary for acclimation to the degree of cold experienced. OF MODELING

*

02.

c 0.2

mRNA

RESULTS

EXPRESSION

ANALYSIS

Control?

As pointed out above, it seems evident that for glycerol-phosphate dehydrogenase, the level of activity (the protein level) is fully regulated at the pretranslational level (Fig. 5A). However, at first sight (Fig. 5B) it would not seem that the thermogenin level could be pretranslationally determined, i.e., determined by the level of thermogenin mRNA (which in its turn would be determined by transcriptional activity and mRNA halflife). We have, however, attempted to analyze the data further, particularly by describing a simple mathematical model for pretranslational/translational control of gene expression and by using this model for computerized analysis of the experimental data by a reiterative simulation procedure (described below). Our aim was to examine if, even in the case of thermogenin, a plausible mathematical solution existed that did not involve a translationally or posttranslationally regulated step. The model was based on the fact that in a pretranslationally regulated system, the stability (here referred to as the degradation constant 8) of the protein product influence&he characteristics of the transient phase (10) (principally the same reasoning applies to pharmacokinetics). The model used was the one depicted in Fig. 6. The purpose was thus to investigate if a value for 8 existed that would allow for explanation of the relationship between mRNA and protein levels within this simple model, implying only pretranslational control. The model as such allows for solely pretranslational control as well as for additional translational control: i.e., the p value can be varied independently of S: However, to examine if a mathematical solution existed that did not involve a translationally or posttranslationally regulated step, p was not varied independently of 6.

0.8

1.0

to maximal)

0.0

0.2

0.4

0.6

0.8

1.0

mRNA level (relative to maximal)

Rather, the rate of protein degradation 6 was systematically varied to simulate different protein half-lives, and the-proportionality constant p determining protein synthesis based on mRNA levels was adjusted for each 8 value to obtain initial steady-state levels of protein. The input function level of mRNA (R,) was then altered in accordance with the experimentally observed time dependence of this parameter, and the size of the protein pool at any point in time then calculated as P,,, = P, &P, + p*Rt. For glycerol-3-phosphate dehydrogenase, the input function R, (i.e., the level of mRNA) was the exponential function found to best fit the data presented in Fig. 3B (see above) (shown as the dotted line in Fig. 7A). The other curves shown in Fig. 7A are examples of simulation results obtained when 8 was varied; here results obtained with 8 values corresponding to protein halflives of 0.1 or 10 days are displayed. The activities experimentally measured (from Fig. 30) are shown as points. It is seen that the data points reasonably fit the

Fig. 6. A sketch of the model used to simulate the transition events. This model is a general one, but as used here, the restrictions were as follows. The mRNA pool R was a function of time Rt, with values obtained from the stated exponential equations derived from experimental observations of the mRNA levels. In the model, both the translation efficiency p and the degradation proportionality factor 6 can be varied, but in the analysis attempted here, the degradation proportionality factor 8 was the only independent variable. The value of the proportionality translation efficiency factor p (for protein synthesis as an effect of mRNA level) was adjusted for each analyzed value of 6 to allow for steady-state protein pools P under initial conditions but was then kept constant. The size of the protein pool (or enzyme activity pool) Pt was calculated by Euler’s rectangular method for -numerical integration (3) and was defined as a function of the initial size (PO), the mRNA (R,), and the proportionality factors p and 6 i e p = Pt - 6*Pt pool + P*Rt. For the present simulations, the t+1 ’ initial levels Rn and Pn were both set to 1. The time resolution t,” used for the reiterative analysis was 0.1 day.

THERMOGENIN

GENE

R1005

EXPRESSION

200 “x 100

0.1

1

10

0.1

Time in cold (days)

10

GPDH haklife

(days)

30-

“n

20-

Fig. 7. Simulation of the transition events based on the model shown in Fig. 6 and calculated by the reiterative method detailed in Fig. 6 legend. For GPDH (A), as well as for thermogenin (C), the mRNA level Rt was set as a function of time in the cold; the equations used were those derived (see text) from the data in Figs. 3B and 4B; the input value of mRNA is indicated by the dotted line. Other curves illustrate results of simulations with S values corresponding to the indicated protein half-lives. Points indicate the experimental data measured (from Figs. 30 and 4D), with initial values adjusted to 1. It may be observed that for both systems the rise phase is slightly more steep than that predicted by the model. In B and D, merit curves for the goodness of fit of simulation results with varied values for protein half-life are shown; the goodness of fit is expressed as the x2 value, calculated as explained in the text.

10-

0

1

1

Time in cold (days)

I

3

Thermogenin

simulated 0.Lday curve but not the lo-day curve. In Fig. 7B we show the merit curve for the fit between the model prediction and the real data as a function of glycerol-phosphate dehydrogenase half-life. The merit curve was calculated as the x2 value obtained (24) from the formula x2 = Z[(P; - P(t,6)/ai) for all 7 time points (; equals 1 to 7), where Pi is the real value at each time point, P(t,8) is the value calculated at the time t as a function of each tested degradation factor 6, and CQhas been estimated as the standard deviation of each data point (indicated on Fig. 30). It is seen that the goodness of fit increases with decreasing values for glycerolphosphate dehydrogenase half-life tested, down to = 1 day, and then plateaus. Thus although the x2 values are somewhat high, it can still be concluded that the data would reasonably fit the constraints of a pretranslationally regulated system, provided that the half-life of glycerol-phosphate dehydrogenase was 1 day or shorter. For longer half-lives, the model becomes very unlikely. We are, however, not aware of any information on the real half-life of glycerol-phosphate dehydrogenase. For thermogenin, the input function R, used (i.e., the level of mRNA) was the exponential function found to best fit the data presented in Fig. 4B (see above). The curves shown in Fig. 7C are examples of results obtained when S was varied; here 8 corresponding to protein half-lives of 2, 5, or 10 days are displayed. The activities experimentally measured (from Fig. 40) are shown as points. The large temporal difference between the changes in mRNA level (dotted line) and the increase in the total amount of thermogenin (points) is very evident. It can, however, be seen that it was possible to identify values for the degradation factor 6 that resulted in a good fit between the thermogenin amount predicted

5

10

9

20

half-life (days)

by the model and the data points obtained; indeed, especially concerning the apparent time delay, the real values excellently fit the curve predicted from a thermogenin half-life of 5 days. In Fig. 7B we show the merit curve for the fit between the model prediction and the real data as a function of thermogenin half-life. It is seen that the x2 value has a clear minimum between 4 and 8 days. Because x2 values below 13 (for 6 degrees of freedom) are nonsignificant, it can even be established within this statistical analysis that there is no statistical reason to reject the model for thermogenin half-lives of 4-8 days (as discussed in Ref. 24, this is a rare situation in model-fitting statistics). Thus, although a cursory look at the correlation between thermogenin mRNA and thermogenin content (Fig. 5B) would suggest that thermogenin amount was not under pretranslational control, we have here demonstrated that a mathematical solution to the problem exists, indicating that the regulation of thermogenin amounts can be explained by a purely pretranslationally controlled model. A practical evaluation of the model demands information about the actual half-life of thermogenin: would it fit with the constraints of the model calculated here, i.e., is it 4-8 days? Some estimates of thermogenin half-life have previously been published; the half-lives observed were in the order of 3-7 days (9,17,23,25). We would especially like to point out that an experiment performed independently of the results presented here, but on the same strain of mice living under identical conditions and with similar methods being used for assessment of thermogenin amount, arrived at an estimate of thermogenin half-life of 7 days (25). Thus there is excellent agree-

R1006

THERMOGENIN

ment between the thermogenin half-life predicted from a pretranslationally regulated model, optimized to fit the data presented here, and the half-life independently measured in vivo. Although this coincidence in itself does not constitute a functional proof, a justifiable conclusion would be the simplest hypothesis: that the thermogenin level is predominantly under pretranslational control. Thus we find that the final thermogenin level is probably determined by the thermogenin mRNA level, but it should be emphasized that in cases like this, with an extremely long transition phase, misleading conclusions concerning regulation of gene expression may be drawn if only non-steady-state conditions are examined. The large differences between the implicated halflives of glycerol-phosphate dehydrogenase and thermogenin may be suggested to be associated with the -disparate localization of the two proteins in the cell, viz. the cytosol and the mitochondria. Mitochondrial turnover seems to be in the order of 7-14 days (5). Conclusions In the present investigation we have shown that for both the recruited enzymes investigated (glycerolphosphate dehydrogenase and thermogenin) the level of mRNA determines the final level of the corresponding protein. However, the transition phase (i.e., the time delay between the attainment of the induced level of mRNA and of the corresponding protein) was very different for the different proteins, apparently due to differences in the protein half-lives, which in turn may be associated with the different intracellular localizations of the proteins. We have also noted that the animal s, apparently without function .a1 feedback, could adjust the thermogenin mRNA level to that needed to determine a level of thermogenin that is physiologically adequate, weeks before this level of the protein was reached. We thank Barbro Svensson for technical assistance, Dr. L. P. Kozak for providing the glycerol-phosphate dehydrogenase cDNA clone, and Professor Anders Martin-Lof for valuable discussions. This investigation was supported by a grant from the Swedish Natural Science Research Council. M. Mtihleisen was a visiting scientist from Biochemisches Institut der Universitat, Im Neuenheimer Feld 328, D-6900 Heidelberg, Germany. Address for reprint requests: J. Nedergaard, The Wenner-Gren Institute, The Arrhenius Laboratories F3, Stockholm Univ., S-106 91 Stockholm, Sweden. Received

9 November

1992;

accepted

in final

form

13 April

1994.

REFERENCES 1. Ashwell, M., G. Jennings, D. Richard, D. M. Stirling, and P. Trayhurn. Effect of acclimation temperature on the concentration of the mitochondrial “uncoupling” protein measured by radioimmunoassay in mouse brown adipose tissue. FEBS Lett. 161: 108-112,1983. 2. Bouillaud, F., D. Ricquier, J. Thibault, and J. Weissenbath. Molecular approach to thermogenesis in brown adipose tissue: cDNA cloning of the mitochondrial uncoupling protein. Proc. Natl. Acad. Sci. USA 82: 445-448, 1985. 3. Bowker, D. W. Dynamic models of homogeneous systems. In: Biological Data Analysis. A Practical Approach, edited by J. C. Fry. Oxford: IRL, 1993, p. 313-343.

GENE

EXPRESSION

4. Bradford, M. M. A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal. Chem. 72: 248-254,1976. 5. Bukowiecki, L., and J. Himms-Hagen. Decreased half-life of some mitochondrial proteins in skeletal muscle and brown adipose tissue of cold-acclimated rats. Can. J. Physiol. Pharmacol. 49: 1015-1018,197l. 6. Cannon, B., A. Hedin, and J. Nedergaard. Exclusive occurrence of thermogenin antigen in brown adipose tissue. FEBS Lett. 150: 129-132,1982. 7. Cannon, B., J. Nedergaard, and U. Sundin. Thermogenesis, brown fat and thermogenin. In: SuruiuaZ in the Cold, edited by X. J. Musacchia and L. Jansky. Amsterdam: Elsevier-North, 1981, p. 99-120. 8. Desautels, M. Mitochondrial thermogenin content is unchanged during atrophy of BAT of fasting mice. Am. J. Physiol. 249 (Endocrinol. Metab. 12): E99-El06,1985. 9. Desautels, M., R. A. Dulos, and B. Mozaffari. Selective loss of uncoupling protein from mitochondria of surgically denervated brown adipose tissue of cold-acclimated mice. Biochem. CeZZ BioZ. 64: 1125-l 134,1986. 10. Hargrove, J. L., and F. H. Schmidt. The role of mRNA and protein stability in gene expression. FASEB J. 3: 2360-2370, 1989. 11. Herron, D., M. N&had, S. Rehnmark, B. D. Nelson, J. Nedergaard, and B. Cannon. Effects of cholera toxin on gene expression in brown preadipocytes differentiating in culture. Am. J. Physiol. 257 (Cell Physiol. 26): C920-C925, 1989. 12. Houstek, J., B. Cannon, and 0. Lindberg. Glycerol-3phosphate shuttle and its function in intermediary metabolism of hamster brown-adipose tissue. Eur. J. Biochem. 54: 11-18, 1975. 13. Jacobsson, A., U. Stadler, M. A. Glotzer, and L. P. Kozak. Mitochondrial uncoupling protein from mouse brown fat: molecular cloning, genetic mapping, and mRNA expression. J. BioZ. Chem. 260: 16250-16254,1985. 14. Kozak, L. P., and E. H. Birkenmeier. Mouse sn-glycerol-3phosphate dehydrogenase: molecular cloning and genetic mapping of a cDNA sequence. Proc. Natl. Acad. Sci. USA 80: 3020-3024,1983. 15. Kozak, L. P., and J. T. Jensen. Genetic and developmental control of multiple forms of L-glycerol 3-phosphate dehydrogenase. J. BioZ. Chem. 249: 7775-7781, 1974. 16. Lin, C. S., and M. Klingenberg. Isolation of the uncoupling protein from brown adipose tissue mitochondria. FEBS Lett. 113: 299-303,198O. 17. Milner, R. E., and P. Trayhurn. Cold-induced changes in uncoupling protein and GDP binding sites in brown fat of ob/ob mice. Am, J. Physiol. 257 (Regulatory Integrative Comp. Physiol. 26): R292-R299,1989. 18. Nedergaard, J., and B. Cannon. [3H]GDP binding and thermogenin amount in brown adipose tissue mitochondria from coldexposed rats. Am. J. PhysioZ. 248 (CeZZ Physiol. 17): C365-C371, 1985. 19. Nedergaard, J., and 0. Lindberg. The brown fat cell. Int. Rev. Cytol. 74: 187-286, 1982. 20. Nichoils, D. G., S. A. Cunningham, and E. Rial. The bioenergetic mechanisms of brown adipose tissue thermogenesis. In: Brown Adipose Tissue, edited by P. Trayhurn and D. G. Nicholls. London: Arnold, 1986, p. 52-85. 21. Nicholls, D. G., and R. M. Locke. Thermogenic mechanisms in brown fat. Physiol. Rev. 64: l-64, 1984. 22. Oufara, S., H. Barre, J. Rouanet, and J. Chatonnet. Adaptation to extreme ambient temperatures in cold-acclimated gerbils and mice. Am. J. Physiol. 253 (Regulatory Integrative Comp. Physiol. 22): R39-R45, 1987. 23. Peachey, T., R. R. French, and D. A. York. Regulation of GDP binding and uncoupling-protein concentration in brown-adiposetissue mitochondria. Biochem. J. 249: 451-457,1988. 24. Press, W. H., S. A. Teukolsky, W. T. Vetterling, and B. P. Flannery. NumericaL Recipes in Fortran. The Art of Scientific Computing (2nd ed.). New York: Cambridge Univ. Press, 1992. 25. Puigserver, P., D. Herron, M. Gianotti, A. Palou, B. Cannon, and J. Nedergaard. Induction and degradation of the uncoupling protein thermogenin in brown adipocytes in-vitro and

THERMOGENIN in-vivo. Evidence for a rapidly degradable pool. Biochem. J. 284: 393-398,1992. 26 Ratner, P. L., M. Fischer, D. Burkart, J. R. Cook, and L. P. Kozak. The role of mRNA levels and cellular localization in controlling sn-glycerol-3-phosphate dehydrogenase expression in tissues of the mouse. J. BioZ. Chem. 256: 3576-3579, 1981. 27 Ricquier, D., F. Bouillaud, P. Toumelin, G. Mory, R. Bazin, J. Arch, and L. Penicaud. Expression of uncoupling protein mRNA in thermogenic or weakly thermogenic brown adipose tissue. Evidence for a rapid P-adrenoceptor-mediated and transcriptionally regulated step during activation of thermogenesis. J. BioZ. Chem. 261: 13905-13910,1986. 28. Ricquier, D., S. Raimbault, 0. Champigny, B. Miroux, and F. Bouillaud. The uncoupling protein is not expressed in rat liver (Comment). FEBS Lett. 303: 103-106,1992. 29. Ricquier, D., J. Thibault, F. Bouillard, and Y. Kuster. Molecular approach to thermogenesis in brown adipose tissue. Cell-free translation of mRNA and characterization of the mito-

GENE

30.

31

32

33

EXPRESSION

R1007

chondrial uncoupling protein. J. BioZ. Chem. 258: 6675-6677, 1983. Spiegelman, B. M., and S. R. Farmer. Decreases in tubulin and actin gene expression prior to morphological differentiation of 3T3 adipocytes. CeZZ 29: 53-60, 1982. Sundin, U., G. Moore, J. Nedergaard, and B. Cannon. Thermogenin amount and activity in hamster brown fat mitochondria: effect of cold acclimation. Am. J. Physiol. 252 (Regulatory Integrative Comp. PhysioZ. 21): R822-R832, 1987. Svensson, B. C., J. Nedergaard, S. Alexson, and B. Cannon. Immunological determination of thermogenin content in brown adipose tissue: validation of the method. In: Thermoregulation: Research and CZinicaZ Applications, edited by P. Lomax and E. Schonbaum. Basel: Karger, 1989, p. 130-133. Thurlby, P., and P. Trayhurn. Regional blood flow in genetically obese (ob/ob) mice. The importance of brown adipose tissue to the reduced energy expenditure on non-shivering thermogenesis. Pfluegers Arch. 385: 193-201, 1980.